Two USRPS but more errors

71 views
Skip to first unread message

Kevin McDermott

unread,
Sep 4, 2013, 3:43:17 PM9/4/13
to flex...@googlegroups.com, vuk.ma...@gmail.com

Fellas,

 

I now have two usrps up and running on two different computers.  However, am back to getting a stream of straight  “U”s when running the TX_USRP on either of the machines. When running the RX_USRP on the desktop (first install) I get a stream of straight “D”s.  When running the RX_USRP on the desktop (second install) I get a stream of straight “0”s.

 

Both configs are using two cores, with time_slot_ns = 1200000

 

I await your response and hope that I can get this going soon! So close to having the USRPs seeing each other!

 

Thanks again as always,

Kevin

 

 

 

 

Ismael Gomez

unread,
Sep 6, 2013, 12:23:34 PM9/6/13
to flex...@googlegroups.com

Hi,

What do you mean with first or second install?

The screen output "O" means overflows and is the receive equivalent to the transmiter's underflow. That means you should follow the same procedure as before.

First try to tx and rx to/from a file with a timeslot of 1 ms. Choose the cores appropiately. Make sure you run as root and keep the background tasks to a minimum. The play with the blocking option in the app file and reduce the time slot length.

Also, let mw know in detail what steps did you follow and what kind of response did you obtain.

--
You received this message because you are subscribed to the Google Groups "ALOE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flexnets+u...@googlegroups.com.
To post to this group, send email to flex...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/flexnets/006601cea9a7%2400d36690%24027a33b0%24%40hume.ictas.vt.edu.
For more options, visit https://groups.google.com/groups/opt_out.

Kevin McDermott

unread,
Sep 6, 2013, 12:29:12 PM9/6/13
to flex...@googlegroups.com

Ismael,

 

Sorry I had a grammar error there. 1st install meant is on my desktop (i7) and 2nd install is on my *laptop* (i5). So I have an install of ALOE on each machine. 

 

I will get to changing the parameters and let you know what happens!

 

Thank you,

Kevin

Kevin McDermott

unread,
Sep 6, 2013, 4:24:04 PM9/6/13
to flex...@googlegroups.com

Ismael,

 

I have run the to/from (tx&rx) file.  It works every time.  I have the timeslot set to 1ms and blocking = 0.  I let the receiver work for about a minute and got an increasing number of blocks while Total Errors and BER remained 0. Around 41,400 blocks I started getting errors and a small BER. The errors increased from there. When I ran the tx/rx file on my laptop I let the tx file run for a longer period of time and errors in the RX did not start occurring until 86400 total blocks.  So, I assuming that errors are occurring when there are no more packets to be analyzed/received. 

So, tx/rx_file work fine on both of my machines.

 

When running the USRPs I have been manipulating the time_slot_ns value. While changing the timeslot for the TX, if the value is too low I will get this:

 

                “[error at sink, ts=12272]: sent 1452/1920 samples

                Module sink returned error from work() function. Trying a clean stop ……”

 

If the timeslot is to high I get the Us for underloaded.

 

 

I have the RX_USRP file running on my laptop, timeslot = 1ms. Blocking = 0, I get an “O” every 25s or so.  But it is running I assume.

 

I will continue to change the blocking on the TX and see what happens. In the event that I get both running without any “u”s or “O”s, what output should I be looking for?

 

Thank you,

Have a great weekend,

Kevin

From: flex...@googlegroups.com [mailto:flex...@googlegroups.com] On Behalf Of Ismael Gomez
Sent: Friday, September 06, 2013 12:24 PM
To: flex...@googlegroups.com
Subject: Re: [ALOE] Two USRPS but more errors

 

Hi,

Ismael Gomez

unread,
Sep 16, 2013, 11:37:51 AM9/16/13
to flex...@googlegroups.com
Hi kevin, sorry for the late answer. 


2013/9/6 Kevin McDermott <kevi...@hume.ictas.vt.edu>

Ismael,

 

I have run the to/from (tx&rx) file.  It works every time.  I have the timeslot set to 1ms and blocking = 0.  I let the receiver work for about a minute and got an increasing number of blocks while Total Errors and BER remained 0. Around 41,400 blocks I started getting errors and a small BER. The errors increased from there. When I ran the tx/rx file on my laptop I let the tx file run for a longer period of time and errors in the RX did not start occurring until 86400 total blocks.  So, I assuming that errors are occurring when there are no more packets to be analyzed/received. 

So, tx/rx_file work fine on both of my machines.



Yes, this is normal, it means there are no more packets to be analyzed/received. 
 

 

When running the USRPs I have been manipulating the time_slot_ns value. While changing the timeslot for the TX, if the value is too low I will get this:

 

                “[error at sink, ts=12272]: sent 1452/1920 samples

                Module sink returned error from work() function. Trying a clean stop ……”

When using the USRP, since you are not using a hard real-time system, you should use the "blocking=1" option in the sink and rx module of the osld_usrp_tx.app and osld_usrp_rx.app files, respectively. The timeslot must be around 900 us. 
 

 

If the timeslot is to high I get the Us for underloaded.


Yes, this is normal. The rule is: 

- If you use a hard real-time system, the timeslot must be exactly 1920/SR (SR: sampling rate)
- Otherwise, the time slot must be *shorter* than 1920/SR
 
When aloe starts, it outputs the sampling rate if the USRP can't work at exactly 1.92 MHz (i.e. if you don't see any message, SR=1.92 MHz) 

 I have the RX_USRP file running on my laptop, timeslot = 1ms. Blocking = 0, I get an “O” every 25s or so.  But it is running I assume.

 

If you have blocking=0, the usrp buffer eventually becomes full. This is normal, but yes, it is working. 
 

I will continue to change the blocking on the TX and see what happens. In the event that I get both running without any “u”s or “O”s, what output should I be looking for?


I guess that now your tx is working. Could you verify this with an spectrum analyzer? For instance, you can use gnuradio's uhd_fft.py. 
 
The next step is to calibrate the receiver gain. Unfortunately, the automatic gain control (AGC) module is not working. It's in the TODO list. The calibration procedure is simple, though: 

1) cmake the source with 'cmake ../ -DLOG=ON' (this is the default command described in the wiki)

2) Create a directory named 'reports/' in the directory where you run the 'runcf' application

3) Run the receiver, as usual. 

4) When it finishes, you should see an output like ... Writting to log file ./reports/synchro_pss.log...done ...

5) open the reports/synchro_pss.log file. You should see something like 

               "Peak detected 11037.5 at 960, cfo=0.01"

    every 10 timeslots. 

   5a) If you don't see this output, increase "scale" parameter in the 'agc' module in the osld_rx_usrp.app file. 
   5b) If the message is written more frequently than every 10 timeslots, reduce the "scale" parameter in the 'agc' module

The scale parameter is the receiver gain. The PSS synchronization computes the correlation of the PSS signal and begins the frame/freq synchronization if the correlation peak is higher than 10000. Thus, you have to tune the receiver gain so that the correlation peak exceeds 10000 every frame (i.e. every 10 time slots, or 10 ms). 

Regards,
Ismael



Kevin McDermott

unread,
Sep 18, 2013, 3:14:49 PM9/18/13
to flex...@googlegroups.com

Ismael

 

Good news! With blocking all on and the timeslot set to 870ms it works! I actually trigger a response from the receiver. I verified that the transmitter is working with a spectrum analyzer as well.  However, I am not in the same position as Ofir is. 

 

I am continually letting the transmitter run.  Then will run (‘r’) the receiver and it immediately kicks back saying

 

Total blocks: 1 Total errors: 1 BLER=1

Total blocks: 1 Total errors: 1 BLER=1

Total blocks: 1 Total errors: 1 BLER=1

 

So they do see each other.  Is the next move to adjust the AGC ? Ofir and I are in the same spot I believe.

 

Thanks

 

As always,

Kevin

 

From: flex...@googlegroups.com [mailto:flex...@googlegroups.com] On Behalf Of Ismael Gomez
Sent: Monday, September 16, 2013 11:38 AM
To: flex...@googlegroups.com
Subject: Re: [ALOE] Two USRPS but more errors

 

Hi kevin, sorry for the late answer. 

--

You received this message because you are subscribed to the Google Groups "ALOE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flexnets+u...@googlegroups.com.
To post to this group, send email to flex...@googlegroups.com.

Ismael Gomez

unread,
Sep 18, 2013, 5:00:25 PM9/18/13
to flex...@googlegroups.com

Glad to here these good news!

Yes, the next step is to manually adjust the agc as i explained to ofir.

Regards

Kevin McDermott

unread,
Sep 20, 2013, 3:08:13 PM9/20/13
to flex...@googlegroups.com

Ismael,

 

I have followed your directions to the letter, but am having some issue it seems with the logging.  I ran the “cmake..” and then made the “reports” directory within the “rtdal_lnx” directory.  I ran the receiver and it stopped again immediately after with the same output as in my last email.  There is no output saying “Writing to log file ./reports/synchro_pss.log...done”.  After I hit CTRL+C to exit, I do however get this:

 

“[rtdal_log_flush():106]: Error opening file ./reports/ctrl.log: fopen: No such file or directory

[rtdal_log_flush():106]: Error opening file ./reports/synchro_pss.log: fopen: No such file or directory

|….

|….

|….”

 

…… with more of the same errors but different directories that should apparently be there.

 

Not sure what is going on here and why the directories aren’t being made. Ideas?

 

Thanks,

Kevin

 

From: flex...@googlegroups.com [mailto:flex...@googlegroups.com] On Behalf Of Ismael Gomez


Sent: Monday, September 16, 2013 11:38 AM

To: flex...@googlegroups.com
Subject: Re: [ALOE] Two USRPS but more errors

 

Hi kevin, sorry for the late answer. 

--

You received this message because you are subscribed to the Google Groups "ALOE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flexnets+u...@googlegroups.com.
To post to this group, send email to flex...@googlegroups.com.

Ismael Gomez

unread,
Sep 22, 2013, 2:36:31 PM9/22/13
to flex...@googlegroups.com
hi, 

The directory "reports" must be created on the same directory where you run the runcf command, e.g. in app_dl/ (not in rtdal_lnx)


2013/9/20 Kevin McDermott <kevi...@hume.ictas.vt.edu>
Reply all
Reply to author
Forward
0 new messages